While the trend to use more and more design intellectual
property (IP) has considerably reduced design effort per gate, it has had the
exact inverse effect on the functional verification effort. In fact, since
integrating multiple design IP blocks is now the norm, verification has
become the dominant task and source of risk in system-on-chip (SOC) projects.
To cope with this challenge and contain these risks, design and verification
teams recognize they need help.
The majority of design and verification teams today use
verification IP (VIP) in one form or another. In fact, VIP is essential for
any complex protocol or bus standard. Teams achieving the greatest success
combine VIP use with a verification reuse strategy and a methodology that
begins with an initial plan and goes all the way through to full verification
closure.
Further contributing to verification IP's ascendance is the
need to ensure compliance with the complex protocol's checklists. This is a
major undertaking in itself and also must span the full verification process.
VIP's end goal is to improve the verification process while reducing your
verification project's risk profile. However, the decision as to which VIP to
use has only gotten more complicated. Just as automobiles range from stripped
down to high end models, several VIP classes now exist.
Since VIP is not created equal, it is critical to select
not just any VIP, but the right VIP that balances your needs and resources
against risks. When the wrong VIP is selected teams commonly face project
delays and even outright project cancellations. This article will help you
optimize VIP selection decisions. Furthermore, it will familiarize you with
the various classes of VIP and the key issues to consider in your VIP
selection.
Key
considerations in VIP selection
Since different classes of VIP vary significantly, in
functionality it is essential to know what capabilities you need. Start the
selection process by identifying the critical requirements for the
verification project. Listed below are the most common questions to consider.
- Will this verification effort be a "quick-and-dirty"
integration where you trust the blocks, or will you perform a complete
functionality verification?
- How will you reuse VIP when moving from block to chip or
to system level verification?
- How will you ensure compliance for any complex
protocols?
- Will you simultaneously verify multiple protocols?
- How will you specify your verification goals and measure
verification progress?
How you answer these questions narrows the range of
applicable VIP. For example, "quick-and-dirty" integration requires far less
capable VIP than a complete verification. Simple integration can be done with
a Bus Functional Model (BFM). Somewhat more complex projects can succeed with
"Testbench VIP."
On the other hand, IP development and SOC projects place
far greater stress on the VIP. They need End-to-End VIP that provides a plan
and clear metrics for measuring completeness and a compliance solution and
that spans the full verification process from architectural modeling to block
design and on to full chip/system verification. It also must span the full
set of verification engines including formal analysis, simulation,
acceleration and emulation.
It is also important to consider your future needs. Even if
you don't need many of these capabilities now, it often makes sense to select
VIP with headroom to retain your investment as your needs grow.
Matching your needs relative to VIP capabilities
There are three primary classes of VIP. They are BFMs,
"Testbench VIP," and "End-to-End VIP." Each class of VIP has its own features
and tradeoffs. They are characterized in Figure 1 below.

Figure 1 — Characterizing VIP alternatives
BFMs and Testbench VIP will serve the need for basic block
connectivity checkout. For anything more complex, End-to-End VIP is needed.
While a variety of suppliers offer commercial VIP, you'll find that most
semiconductor and systems companies internally develop only BFMs.
Page 2:
Verification IP takes a broader role
Page 3:
Verification IP takes a broader role